Agronomy Journal Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 17 August 2005
Published in Agron J 97:1322-1332 (2005)
DOI: 10.2134/agronj2005.0008
© 2005 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ball Coelho, B. R.
Right arrow Articles by Bruin, A. J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Ball Coelho, B. R.
Right arrow Articles by Bruin, A. J.
Agricola
Right arrow Articles by Ball Coelho, B. R.
Right arrow Articles by Bruin, A. J.
Related Collections
Right arrow Soil Fertility and Productivity
Right arrow Animal Waste
Right arrow Best Management Practices
Right arrow Maize Management

Production Papers

Optimization of Liquid Swine Manure Sidedress Rate and Method for Grain Corn

B. R. Ball Coelhoa,*, R. C. Royb and A. J. Bruina

a Agric. & Agri-Food Canada, Southern Crop Protection & Food Res. Cent., 1391 Sandford St., London, ON, Canada N5V 4T3
b Agric. & Agri-Food Canada, Southern Crop Protection & Food Res. Cent., Delhi, ON, Canada, N4B 2W9

* Corresponding author (ballb{at}agr.gc.ca)

Received for publication January 6, 2005.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 
Sidedressing may provide a better window of opportunity for land application of liquid swine (Sus scrofa) manure than early spring or fall application. Rates could be fine-tuned to match crop N demand using the presidedress nitrate test (PSNT) if: (i) the yield response function to sidedress rate is consistent and (ii) yield and PSNT are positively correlated. To optimize application rate and method, we measured corn (Zea mays L.) grain yield response to in-row injection (INJ) and topdress (TD) of liquid swine manure (LSM) sidedressed at different rates on clay loam (51-cm rows in 1999) and silt loam (75-cm rows from 2000–2002). Yields exceeded local long-term averages with INJ in all but the wettest year, were variable with TD, and were 2 Mg ha–1 greater with INJ than TD at 37.4 m3 LSM ha–1. From the quadratic yield response to sidedress injection rate, optimal rate (to achieve 95% maximum yield) ranged from 38 to 63 m3 ha–1 (plot-scale data; four 6-m sections per plot) and 37 to 49 m3 ha–1 (field-scale data; 0.2-ha plots). Yields were correlated with the PSNT (r = 0.75 for no LSM sidedress; r = 0.24 for all treatments). Given the consistent yield response to sidedress INJ rate and accurate (correct 88% of the time) PSNT-based predictions of additional N requirements (from comparisons of N fertilizer recommendation and relative yield), sidedress injection of LSM using the PSNT to fine-tune rates according to crop N requirements can be considered as a best management practice.

Abbreviations: bicarb-P, sodium bicarbonate extractable phosphorus • CEC, cation exchange capacity • INJ, inject • LSM, liquid swine manure • Ninorg, inorganic nitrogen • PSNT, presidedress nitrate test • SOM, soil organic matter • TD, topdress • UAN, urea ammonium nitrate


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 
INCREASING NUTRIENT USE EFFICIENCY of land-applied manure, possibly through better placement and timing, could reduce supplemental fertilizer requirements, improve profitability, and minimize environmental impacts. Sidedressing (mid-June) provides a good window of opportunity for application of liquid manure from swine operations for several reasons. At sidedressing time, there is uptake by developing roots, and soil water content is typically lower than in fall or spring. As a result, the potential for soil compaction is reduced, and drainage tiles are less likely to be flowing. Active roots coupled with minimal tile flow can lead to increased nutrient uptake and decreased transfer of material to ground or surface waters. Contaminant movement to surface waters is particularly a concern on tiled land (Dean and Foran, 1992; Shipitalo and Gibbs, 2000). Furthermore, application rates for corn can be determined more precisely at sidedress than in fall or spring because potential exists to use the PSNT [more accurate than the preplant nitrate test (Grove, 1992; Ball-Coelho et al., 2004)] or other tools such as on-the-go sensors (Scotford et al., 1999; Stombaugh and Shearer, 2000) and images (Schnug et al., 1998) to adjust sidedress rates on a site-specific basis according to soil properties or crop response. In Quebec, sidedressing was found to be more economically and agronomically sound than fall spreading (Cote et al., 1999).

The variable-rate manure application method is relatively new, and tools have not yet been developed on which to base application rates. The PSNT (NO3 concentration in the top 30 cm of soil when corn is 15 to 30 cm tall) is used in a number of U.S. states and in Ontario to refine fertilizer N recommendations for corn (Randall et al., 1999; OMAF, 1999). With adoption of the late PSNT, surface water quality was improved relative to a control subwatershed in a recent study (N requirements were supplied predominantly by fertilizer N in both subwatersheds; Jaynes et al., 2004). Critical PSNT values determined originally for fertilizer rates (20 to 30 mg kg–1) may differ in systems where manure is the primary nutrient source, however, according to results from the U.S. northern Corn Belt (Randall et al., 1999). To use the PSNT to accurately adjust sidedress manure rates to meet crop N requirements, there must be a consistent relation between crop yield and sidedress manure rate that can be described mathematically, as well as a correlation between yield response and the PSNT in manured soils.

Information regarding crop response to sidedressed manure gathered using modern application equipment under field-scale conditions is limited. For example, there are few comparisons of the effects of different application methods on the response to sidedress manure rate. Injection rather than in-lay or TD presents a compromise between damage to roots by injectors and improved nutrient availability due to placement in the root zone and reduced NH3 volatilization. Injection can reduce volatilization by 90% without concurrently increasing denitrification N loss (Dendooven et al., 1998). Surface application of manure without incorporation is further susceptible to runoff losses of NH4 and P (Eghball and Gilley, 1999), especially when rain falls soon after application (Cote et al., 1999). Besides reducing runoff and volatilization, injection has the added benefit of solving application-related odor problems.

Field-scale comparisons are needed to determine crop yield responses across varying soil and landscape conditions due to concerns regarding the validity of scaling-up from small plots (Kachanoski and Fairchild, 1996) and to assess and calibrate rate-determining methodologies. Large-scale plots also allow the use of commercial-sized equipment, which coupled with the potential to integrate over variable soil conditions, increase credibility and practical application since decisions are made and decision support systems generally employed at the field-scale (OMAF, 2003; Wagenet and Hutson, 1996). This study focuses on sidedress time of application due to aforementioned advantages. Our objectives were to: optimize LSM sidedress application rate and method for grain corn using state-of-the-art equipment and assess the potential for using the PSNT to refine sidedress manure rates by relating the yield response to both LSM application rate and PSNT-based fertilizer N recommendations.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 
Cultural Practices and LSM Application
Experiments were conducted on clay loam in 1999 (43°20' N, 81°36' W) and on silt loam from 2000 to 2002 (43°44' N, 81°01' W). These Gray Brown Luvisols (Hapludalf) of the Huron Association are typical in the mixed farming zones of southern Ontario. Each year, eight sidedress manure treatments were arranged in a randomized strip plot design with two replicates per treatment (Table 1). The clay loam site was previously cropped with corn while the silt loam site was cropped with wheat (Triticum aestivum L.) before the experiment and had no known history of manure application. The clay loam site was not tilled while fall plowing and spring secondary tillage were completed each year at the silt loam site. Corn was planted in late April or early May in narrow (1999) or wide (2000–2002) rows (Table 1) with 47 L ha–1 6–26–6 in furrow. Two tipping buckets (TE525, Campbell Scientific Inc., Edmonton, AB) recorded June to October rainfall from 2000 to 2002.


View this table:
[in this window]
[in a new window]
  		Table 1. Treatments, events, and rainfall in experiments on clay loam and silt loam sidedressed with liquid swine manure by injection (INJ) or topdress (TD) at different rates.

 
Applicators with in-tank mixing (Nuhn Industries, Sebringville, ON) and electronic flow control (GreenLea Ag Centre, Mount Elgin, ON) were used to apply manure at rates (Table 1) hereafter denoted by LSM0, LSM18.7, etc. Coupled with GPS, this equipment allowed generation of application maps, providing a convenient method of record keeping and demonstrating due diligence during application. A 26-m3 tanker with tandem duals and 11-row injector was used to apply LSM in 1999 while a 15-m3 tanker with six-row injector was used from 2000 to 2002. During calibration, less than 10% flow variability was observed between injectors on the applicator tool bar. Between-row variability was further reduced by mounting restrictors (Nuhn Industries, Sebringville, ON) on the outflow tubes to increase back pressure. Application of manure was completed within 24 h each year in mid- to late June (Table 1). Manure was sampled on the day of application from several different tanker loads and analyzed for dry matter and nutrients (Table 2). Constant agitation of manure in the lagoons (finishing hog barns with wet/dry feeders) during loading, in the nurse tanks (used in 2001 and 2002 to minimize time required to apply all treatments), and in the applicators ensured that uniform material was provided for all treatment plots in a given year.


View this table:
[in this window]
[in a new window]
  		Table 2. Liquid swine manure composition and amount of N (total and NH4), phosphate, and potash supplied by five sidedress rates each year.

 
At the silt loam site, treatments were applied in 2001 and 2002 to the same plots as in 2000, except that INJ rates were reduced to equal those applied in 1999 (Table 1). In 2002, plots designated for LSM18.7 actually received 28.1 m3 ha–1 due to an improper setting on the flow control. Teeth were 15 to 25 cm aboveground for TD and 10 to 15 cm deep for INJ treatments. To simulate physical effects in the zero-rate plots (spaced out across the design), the applicator tank was filled to two-thirds and pulled along the respective plots, either with teeth above (0 TD) or in the ground (0 INJ), but no manure was applied.

Injectors were constructed using Vibra Shanks (Kongskilde Ltd., Strathroy, ON). In 2000, a coulter was added in front of each injector. The injection tool bar was further modified for 2001 and 2002 by mounting disk hillers behind each injector (evolved into the row-crop model, Nuhn Industries, Sebringville, ON). Modifications were in part based on results of testing various injector systems in the fall of 2000 in a nearby field. Injection of rhodamine red dye showed that night crawler (Lumbricus terrestris) burrows facilitated direct transfer of contaminants to tile drains in these soils, by allowing material to bypass matrix flow through soil. Injection tools with more mixing action reduced the probability of intersection of large volumes of applied material with worm burrows, thereby lessening the risk of contaminant movement to tile drains. The coulter increased soil mixing, and the disk hillers allowed for shallower injection by covering the manure with soil. Shallower injection with coverage served to simultaneously minimize potential movement to tile drains, volatilization, odor, and runoff.

Soil and Plant Sampling
Four subsampling locations per plot located approximately 40 m apart along the center of each 12-row plot at each experimental site were flagged using GPS coordinates. Before imposing treatments in the first year at each site, topsoil (0–20 cm) samples (2-cm-diam. cores) were collected using benchmark or "smart" sampling (20 locations based on zones of variation in soil and topography) at the clay loam site on 16 June 1999 and from each of the four subsample locations per plot at the silt loam site on 6 June 2000 (64 samples, each a composite of 20 cores) to assess spatial variability of selected chemical properties (Table 3). Soil texture (top 20 cm) at each site was determined using the hydrometer method (Sheldrick and Wang, 1993).


View this table:
[in this window]
[in a new window]
  		Table 3. Range of available nutrients and selected chemical properties in topsoil (0–20 cm) at the clay loam and silt loam experimental sites before sidedressing with liquid swine manure.

 
In late May or early June (Table 1) before sidedressing each year, topsoil samples (0–30 cm) were collected for inorganic N (Ninorg) determination (from four subsample locations per plot of one replicate and two subsample locations per plot of the second replicate in 1999 and from four subsample locations per plot in 2000–2002). Each sample was a composite of eight (1999, 2000) or nine (2001, 2002) 2-cm-diam. cores per plot. Potential NO3 leaching to deeper layers of the soil profile was assessed by collecting cores (5.1 cm diam.) from LSM0 INJ, LSM37.4 INJ, LSM74.8 INJ and LSM37.4 TD treatments at three subsample locations per plot about 3 mo after the second (on 27 Aug. and 11 Sept. 2001) and third (4–5 Sept. 2002) application of LSM at the silt loam site. A hydraulically driven soil sampler (Giddings Co., Ft. Collins, CO) mounted on a Model 470 Hi-Tractor (Hagie Manufacturing Co., Clarion, IA) for self-propulsion through standing corn was used to obtain deep soil samples. It was equipped with a zero recess bit, 0.6-m sleeves, and casing to allow collection of multiple cores per sample hole without contamination from upper layers. Soil was sampled to 180 (2001) and 120 (2002) cm deep, and cores were divided manually into five or seven depth increments (0–20, 20–40, 40–60, 60–90, and 90–120 in both 2001 and 2002, as well as 120–150 and 150–180 cm in 2002). All soil was homogenized before extraction by pushing through a 6-mm opening screen, and Ninorg was extracted by shaking 12.5 g of field-moist soil in 25 mL 2 M KCl for 1 h. Concentrations of NO3–N and NH4–N in the filtered extracts (Maynard and Kalyra, 1993) were determined using continuous-flow colorimetry (Tel and Rao, 1981) for samples collected from 1999 to 2001 and flow injection (Lachat Instruments, Milwaukee, WI) colorimetry (Liao, 1999; Diamond, 2001) for 2002 samples. Concentrations were corrected to a dry soil weight basis using gravimetric water content determined separately for each sample.

Before combining, plot-scale yield was determined by manually harvesting corn cobs (Table 1) from a 6-m section of one row in the center of each of the four subsampling locations per plot after counting the number of stalks (and broken stalks in 2000–2002). Manually harvested cobs were shelled using a cob sheller or plot combine and weighed, and from a representative subsample, grain moisture content (using an electronic Dickey-John II gauge) and test weights were measured. To determine yield at the larger field scale, grain from the remainder of the plots was harvested using a combine (Case IH in 1999, 2000, and 2002; John Deere in 2001) equipped with a yield monitor and GPS receiver. Sensors recorded grain yield and moisture content every 1 (2001), 2 (1999–2000), or 3 s (2002) along two six-row passes per plot. Yields (plot and field scale) were corrected to 15.5% moisture using grain moisture determined from the same sample.

In 2000, the south half of the experimental area (included two subsample locations) and the west half of the westernmost plot inadvertently received 145 kg N ha–1 as UAN [urea ammonium nitrate, (NH2)2CO NH4NO3] preplant. Therefore, corn (manual) and PSNT samples were collected only from the unaffected (east) side of the westernmost plot that year. In 2002, 150 kg N ha–1 as urea [(NH2)2CO] was accidentally broadcast pre-emerge on the easternmost plot (LSM74.8 INJ) and the adjacent east six rows of one LSM0 TD plot due to a commercial operator error. Corn and soil samples were thus collected separately from the east and west sides of the LSM0 TD plot to determine fertilizer N response.

Data Analyses
Field- and plot-scale grain yields were compared using PROC CORR (SAS Inst., 1999) by calculating the average of the six combine monitor yield readings nearest to the latitude of the manual sample location (2000–2002). In 1999, the exact location of the four manual samples within each plot could not be pinpointed in field-scale data due to an unreliable GPS signal during combining, so the average of all (30–60) combine monitor yield readings from the respective plot quarter sections were compared to the plot-scale yield measures (Fig. 1) .



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 1. Yearly comparisons of plot- (by manual harvesting) and field- (by combine yield monitor) scale corn grain yield following sidedress of liquid swine manure with different rates and methods. r = correlation coefficient.

 
Variation in topsoil Ninorg and other [soil organic matter (SOM), cation exchange capacity (CEC), pH, sodium bicarbonate extractable P (bicarb-P), Bray-1 P (Olsen and Sommers, 1982), K, Ca, Mg, Zn, Mn, Fe, Cu, B, and S] properties was analyzed statistically using the General Linear Models procedure (SAS Inst., 1999) for analysis of variance (ANOVA) according to the randomized complete block design each year. Topsoil Ninorg data collected from areas where fertilizer N was unintentionally applied and sampling was not modified (e.g., the south two subsample locations in 2000 and the east portion of the LSM0 TD plot in 2002) were excluded from the analysis of treatment effects. Soil profile NO3–N was analyzed using a similar model, with depth as a subplot, in two ways based on the selected treatments from which cores were collected (LSM0, 37.4, 74.8 INJ, and LSM37.4 TD). In the first model, LSM37.4 TD data were excluded to examine the effect of INJ rate (LSM0, 37.4, 74.8) while in the second model, data from LSM0 and LSM74.8 INJ were excluded to examine the effect of method (at LSM37.4).

The additional variability caused by accidental fertilizer N application in 2000 and 2002 was used to examine yield responses to the PSNT. Areas where fertilizer N was accidentally applied were included in two correlation analyses using PROC CORR (SAS Inst., 1999), one between PSNT and corn grain yield in plots receiving no LSM and the other between PSNT and yield in all treatment plots. When grain data from the accidentally fertilized locations were excluded, PSNT remained significantly correlated with plot-scale corn population, grain yield, test weight, and moisture (Table 4) over all treatments and years. Field-scale yield was correlated with the PSNT within each year except 2000 and was not correlated when all (1999–2002) yield data were analyzed together (Table 4).


View this table:
[in this window]
[in a new window]
  		Table 4. Comparison, based on ANOVA, of corn grain yield, quality, and population where liquid swine manure was sidedressed using different methods (inject or topdress) and rates (0, 37.4, or 56.1 m3 ha–1) from 1999 to 2002, and correlation coefficient (r) with preapplication field variability in topsoil NO3–N concentration (PSNT).

 
Since preapplication topsoil NO3–N concentrations varied significantly and were significantly correlated with measured corn parameters, covariate analysis was used to remove PSNT variation effects. Treatment effects on corn population and grain parameters (Table 4) were compared using the Mixed procedure for repeated measures, with year (1999–2002) specified as the repeated effect in a random statement (replicate pooled across method x rate as the subject), subsampling location specified as the residual effect in the repeated statement (year x replicate pooled across method x rate as the subject), and preapplication soil NO3–N concentration as the covariate. To more accurately assess method effects, data collected from LSM18.7 (1999, 2001), LSM28.1 (2002), LSM74.8 (1999–2002), and LSM93.5 (2000) plots were excluded from this analysis (to balance the design). Data from the north end of the experiment (Subsample Locations 1 and 2) of one replicate in 1999 could not be included because corresponding preapplication soil samples were not collected. Data from locations where fertilizer N was inadvertently applied were excluded because PSNT adjustment did not completely remove the fertilizer effect. All data sets were normal. Akaike criteria (Littell et al., 1998) were used to determine the appropriate structure of the covariance matrix as well as whether separate variance estimates each year were required. Models were simplified with respect to the covariate as described in Milliken and Johnson (2002), and when treatment effects varied with the PSNT (grain test weight), means were adjusted using the average PSNT each year and compared within years. Otherwise, when treatment effects were significant, means were compared using the protected LSD at a 0.05 probability level (SAS Inst., 1999).

To quantify grain yield response to injection rate (volume basis) of sidedressed LSM and calculate the optimum LSM injection rate (required for 95% maximum yield), regression curves were fitted each year for modified (fertilizer N locations removed) yield data sets (plot and field scale), both with and without PSNT adjustment. Response equations were not generated for TD since only three rates were tested. Linear and quadratic regression terms were partitioned, and the remaining possible polynomials were pooled into a single term (lack of fit) in each case to ensure that all treatment effects were removed from the variance analysis error term before testing significance of the linear and quadratic components (Bowley, 1999).


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
SUMMARY AND CONCLUSIONS
 REFERENCES
 
Comparison of Field- and Plot-Scale Yield Measures
Plot- and field-scale yields were well correlated each year (Fig. 1) and over all years (r = 0.64, P < 0.0001). Field scale overestimated plot-scale yield at the lower end of the yield range and underestimated yield at the upper end of the range every year except 2002 when field scale exceeded plot-scale yields over the entire range of data and the intercept was substantially larger (6.5 compared with 2.1–3.1 Mg ha–1 in other years, Fig. 1), possibly due to poorer monitor calibration that year. The general trend of overestimation of low yields and underestimation of high yields by monitors could be due to border effects since one outside row of each combine pass would acquire (or lack) nutrients from the adjacent plot. Border effects were avoided during manual harvest by collecting corn from only the center row of each plot.

Soil Ninorg Variation
Presidedress measures of soil Ninorg in the top 30 cm indicated relatively little carryover of N from prior manure applications. Soil NH4 concentrations before sidedress LSM application were low each year (averaged 0.4, 0.8, 1.2, and 0.4 mg kg soil–1 in 1999, 2000, 2001, and 2002, respectively) and unaffected by treatment. Soil NO3 concentrations before sidedressing were greater and more variable than NH4. A small cumulative carryover of N at the highest rate after 2 yr of manure application was indicated by greater PSNT with LSM74.8 INJ (4.4 mg kg–1) than all other treatments (2.9 mg kg–1) at the silt loam site in 2002 (Table 5).


View this table:
[in this window]
[in a new window]
  		Table 5. Comparison, based on ANOVA, of NO3 concentrations in the top 30 cm of soil before liquid swine manure sidedress at five different rates using inject (INJ) or topdress (TD) methods, following 0 (1999 and 2000), 1 (2001), or 2 yr (2002) of previous application.

 
Treatment-independent variation in the PSNT occurred in 1999, 2000, and 2002. Before LSM application in 1999, soil NO3–N concentrations were greater in three plots on one side of the experiment (15–17 mg kg soil–1) than in all other plots (<10 mg kg soil–1), resulting in greater mean NO3–N concentrations for three treatments (15 mg kg–1 for LSM0 TD, LSM18.7 INJ, and LSM37.4 INJ) than for all others (8 mg kg–1, Table 5). The spatial gradient may have been caused by overspread of fertilizer N on the adjacent wheat crop that spring or manure applications in previous years. Areas known to have received fertilizer N in 2000 and 2002 had greater presidedress soil NO3 concentrations. Average soil NO3 concentration in 2000 was nearly three times greater where UAN had been incidentally applied (26 mg kg–1, average of the two south subsample locations, all plots) than in unaffected areas (9 mg kg–1, average of two north subsample locations, all plots). Low soil NH4 concentration at this time (6 June) indicated that N from the preplant UAN application had already converted to NO3 form. When data from fertilized locations were excluded from the analysis, the PSNT was unaffected by treatment in 2000 (statistical model nonsignificant, Table 5). In 2002, soil NO3 concentration was greater in areas unintentionally fertilized with urea pre-emerge (>5 mg kg–1 in the easternmost plot and east half of the adjacent plot) than in unaffected areas. When data from these locations were excluded, all treatment means were <5 mg NO3 kg –1.

Concentrations of soil NO3–N deeper in the profile did not differ with application method after 2 yr (Table 6) but after 3 yr of LSM37.4 averaged greater with INJ (2.8 mg kg–1) than TD (0.4 mg kg–1) over all depths (0 to 120 cm). Soil NO3–N increases with INJ rate after 2 yr (2001) occurred not only in the top 20 cm, but also between 40 and 120 cm deep (Table 6). The absence of rate effects in the 20- to 40-cm layer suggests that the deeper (40–120 cm) rate-dependent pulse likely originated the previous year. In the deepest layers (120–180 cm), concentrations did not vary with application rate, indicating that the pulse (from 2000 application) had not yet moved below the corn root zone. After 3 yr of application (2002), soil NO3–N concentrations were greater where LSM was applied (LSM74.8 and LSM37.4) than in LSM0 at all sampling depths and greater with injection of LSM74.8 than LSM37.4 in the top 20 cm and between 40 and 90 cm deep (Table 6). With LSM37.4 INJ, concentrations were 2.0 mg kg–1 or less below 40-cm depth both years.


View this table:
[in this window]
[in a new window]
  		Table 6. Comparison, based on ANOVA, of application method and injection rate of sidedressed liquid swine manure (LSM) on soil profile NO3–N concentrations after 2 and 3 yr.

 
Corn Response to LSM Sidedress Methods
Corn growth response to LSM application rate and method was highly visible by July (Fig. 2) . Nitrogen deficiency symptoms were noted in the control (LSM0) treatments each year, and TD (LSM37.4 and LSM56.1) corn exhibited greater variability in size and color than INJ. Grain yield response to application method depended on LSM rate (Table 4, Fig. 3) . At the lower rate (LSM37.4), INJ outyielded TD by 2 Mg ha–1 (plot-scale data, Fig. 3). At the higher rate (LSM56.1), yield tended to be greater with INJ than TD, but the difference was smaller (0.5 Mg ha–1, plot-scale data) and not statistically significant. The trend of greater INJ yield advantage with LSM37.4 (1.1 Mg ha–1) than LSM56.1 (0.6 Mg ha–1) was also apparent but not statistically significant for field-scale data (Table 4). With preplant injection, Schmitt et al. (1995) similarly observed greater corn yields (by 0.75 Mg ha–1) than with broadcasting, accompanied by greater soil NO3 concentrations in summer. Method of LSM application did not affect grain quality (Table 4), other than in 2001 when test weights were greater with INJ (68.5 kg hL–1) than TD (66.7 kg hL–1) at LSM56.1.



View larger version (119K):
[in this window]
[in a new window]
  		Fig. 2. Corn growth response to injected (INJ) or topdressed (TD) liquid swine manure sidedressed at different rates (m3 ha–1) on (A) 18 July 2001 and (B) 23 July 2002.

 


View larger version (15K):
[in this window]
[in a new window]
  		Fig. 3. Corn grain yield response to injected (INJ) or topdressed (TD) liquid swine manure (LSM) sidedressed at 0, 37.4, or 56.1 m3 ha–1, average of 1999–2002 plot-scale data. Means with the same letter are not significantly different at the 0.05 probability level.

 
Greater variability and lower yield with TD might be due to reduced nutrient acquisition where LSM was spatially separated from roots, particularly during dry periods, and to N loss via volatilization, which varies with weather and microclimate. Placement of nutrients directly in the root zone and reduced NH3 volatilization apparently compensated for any root or shoot damage during sidedress injection, even in narrow-row corn (1999). Physical disturbance associated with sidedress injection did not damage the crop significantly since yields in the control (LSM0) were equivalent with INJ and TD at both the plot (Fig. 3) and field (Table 4) scale. Furthermore, application method had only minimal effects on corn population (Table 4), with a tendency of fewer plants with INJ (67306 plants ha–1) than TD (68577 plants ha–1, average 1999–2002, P = 0.06). This trend occurred in 3 (1999–2001) of 4 yr. In 2002, corn population (Table 4) was more variable and on average much smaller (CV = 15%, 46 200 plants ha–1) than in previous years (CV = 7%, 75 190 plants ha–1), due mainly to several weeks of cold, wet weather after planting that resulted in soil crusting and poor emergence.

Grain Yield Response to LSM Sidedress Injection Rate
Grain yields with injected manure (Fig. 4) were above the long-term township average of 8 Mg ha–1 for the clay loam site and 8.3 Mg ha–1 for the silt loam site (OMAF, 2003) in all but the abnormally wet 2000 growing season (Table 1) when yield potential overall was reduced. That year, yields ranged from 4.5 to only 9.5 Mg ha–1, whereas in other years, yields ranged from 2.8 to 12.9 Mg ha–1 (Fig. 1). Greater yield with injected LSM (10 Mg ha–1, average of LSM37.4 and LSM56.1) than with 150 kg fertilizer N ha–1 applied pre-emerge (6 Mg ha–1, fertilized half of the LSM0 plot) was observed in 2002 when yield with no fertilizer N or LSM was 3 Mg ha–1. Wall et al. (1997) in southwestern Ontario and Cote et al. (1999) in Quebec found similar yields where corn was sidedressed with either manure or comparable amounts of fertilizer N. Grain and stalk quality were also affected by LSM rate. Grain test weight increased with INJ application rate from 64.0 (LSM0) to 66.2 (LSM37.4) to 67.1 (LSM74.8) kg grain hL–1. Broken stalks were reduced from 1.5 to 0% in 2000, 0.9 to 0.4% in 2001, and 2.6 to 0.4% in 2002 with manure application compared with LSM0, likely due to improved K nutrition with LSM.



View larger version (28K):
[in this window]
[in a new window]
  		Fig. 4. Corn grain yield response curves to sidedress injection rate of liquid swine manure (LSM) from 1999 to 2002. Yields are adjusted for presidedress topsoil NO3 concentration. Large symbols represent the rate required for 95% maximum yield for plot- (colored symbol) and field- (gray symbol) scale data sets. Solid and dashed lines represent quadratic fits for plot- and field-scale yields, respectively.

 
Functions describing grain yield increases with rate of injected LSM (Fig. 4) were improved through adjustment for preapplication soil NO3–N concentration each year, as evidenced by greater coefficients of determination for adjusted (r2 = 0.86) than unadjusted (r2 = 0.61, average of plot- and field-scale data from 1999 to 2002, Table 7) data sets. Greatest adjustment effects occurred in years when preapplication soil NO3–N variability was greatest (r2 increased by 0.4 in both 1999 and 2000, Table 7). Optimum LSM application rate (rate required to achieve 95% maximum yield) was altered by PSNT adjustment only in 1999 for field-scale data, with lower optimum rate derived from adjusted than unadjusted data (Table 7). Fit of the yield–LSM rate relationship was poorer in 1999 (r2 = 0.54, average of unadjusted and adjusted plot- and field-scale data) than in other years (r2 = 0.80, 2000–2002, Table 7), and standard errors for both plot- (0.82 Mg ha–1) and field- (0.11 Mg ha–1) scale yield means were greater in 1999 than in other years (0.25 and 0.07 Mg ha–1, plot- and field-scale measures respectively, 2000–2002 average, Fig. 4). Greater yield variability in 1999 than in other years may have been due to greater soil spatial variability [greater range of values for SOM, CEC, Ca, and Mg (Table 3)] at the clay loam than at the silt loam site.


View this table:
[in this window]
[in a new window]
  		Table 7. Summary of corn grain yield response to sidedress manure injection rate: coefficients of determination (r2) for the relation, 95% maximum yield, manure rate for 95% maximum yield, and P values for the regression partitions for data obtained by manual sampling and from a combine yield monitor, either unadjusted or adjusted for preapplication topsoil NO3–N concentration, 1999–2002.

 
The PSNT-adjusted yield response to injection rate was described reasonably well by a quadratic function (Table 7) each year. Optimum sidedress LSM injection rates (PSNT-adjusted) ranged 38 to 63 m3 ha–1 (plot-scale data, 198–334 kg N ha–1) and 37 to 49 m3 ha–1 (field-scale, Table 7, 196–229 kg N ha–1) from 1999 to 2002 and averaged 50 (plot scale) and 41 (field scale) m3 ha–1. The optimum rate (63 m3 ha–1) calculated from the 2000 plot-scale data likely overestimated LSM requirements, considering it exceeded all other estimates, yields were least in 2000, and the discrepancy between plot- and field-scale-derived estimates was largest that year (24 m3 LSM ha–1 greater for plot- than field-scale data in 2000, Table 7). Correlation between plot- and field-scale yield measures were poorest (r = 0.73, Fig. 1) in 2000, and the flat response to rate caused by wet weather (Fig. 4, Table 7) resulted in reduced accuracy in the calculation of an optimum from quadratic-based formulae. Additional variability in 2000 may have been caused by intense rainfall (60 mm) 9 d after sidedress application, which initiated a runoff event. Excluding the 2000 plot-scale data likely provides a better approximation of optimal LSM injection rate (46 m3 ha–1) and is nearer the field-scale-derived average. Generally, optimum LSM injection rate was less when determined using field- than plot-scale data, except in 2002 due likely to poor monitor calibration that year (Fig. 1). Lower optimum rate estimates from field-scale data could be due to border-induced overestimation of yields in control (LSM0) treatments by the combine monitor (Fig. 1).

Relation among the PSNT, PSNT-based Nitrogen Recommendation, and Yield
Variation in the PSNT caused by both previous manure and inadvertent fertilizer applications provided opportunities to examine yield response to the PSNT. In control (LSM0) plots, a significant positive correlation existed between PSNT and yield, which was greater in 1999 (r = 0.86, P = 0.0004) and 2000 (r = 0.95, P < 0.0001, plot-scale data) when the PSNT range was larger (15–77 kg NO3–N ha–1 in 1999 and 25–183 kg NO3–N ha–1 in 2000) as compared with either 2001 (r = 0.54, P = 0.0315) or 2002 (r = 0.66, P = 0.0014) when PSNT ranges were smaller (12–31 kg NO3–N ha–1 in 2001 and 4–55 kg NO3–N ha–1 in 2002).

To assess response to the PSNT each year, relative yields (=yield/optimum yield that year) were calculated. In control plots in cases when relative yield ≥ 1 (i.e., yield = optimum yield), the corresponding PSNT represented the critical PSNT (no additional N required to optimize yield). Critical PSNT values were 50 kg NO3–N ha–1 in 1999 and 80 kg NO3–N ha–1 in 2000, similar to thresholds above which no inorganic fertilizer is usually recommended (Randall et al., 1999; OMAF, 1999). In 2001 and 2002, PSNT values were not high enough to identify a critical value, as there were no cases in control plots when relative yield ≥ 1.

The PSNT-based fertilizer N recommendation (OMAF, 1999) was compared with relative yield in each plot and the amount of available N applied as LSM. Available N was estimated from NMAN (Nutrient Management Program) software (OMAF, 2003), which assumes that all NH4 (Table 7) plus 25% of organic N is available if injected in season while 66% of NH4–N is available if topdressed on the standing crop. The PSNT-based N recommendation was considered correct if either of two conditions were met: (i) relative yield ≥ 1, and available N applied ≥ amount recommended by the PSNT; or (ii) relative yield < 1, and available N applied < PSNT-recommended amount. Treatments where the amount of N applied was much greater than that recommended were excluded from this comparison. The PSNT-based N recommendation correctly indicated whether additional N was required in 82% of cases (36 of 44) in 1999 and 93% of cases in 2000 (37 of 40). In 2001, in all 32 cases where N applied < PSNT-based recommendation, relative yield < 1, indicating that the PSNT was correct in predicting the additional requirement for N. In 2002, of 52 cases where N recommended > N applied, the PSNT was correct 76% of the time in predicting that more N was required (since relative yield < 1). Most of the cases (10 out of 12) where the PSNT-based recommendation was incorrect (additional N recommended but relative yield = 1) in 2002 occurred in INJ LSM28.1 and LSM37.4, indicating either greater N availability or efficiency than expected in these treatments or some cumulative residual effect after 3 yr.

Short-Term Effects of LSM Application on Soil Test Phosphorus and Potassium
After 3 yr, bicarb-P increased by 5 and 10 mg kg–1, and Bray-1 P increased by 15 and 30 mg kg–1 with LSM56.1 and LSM74.8, respectively. The amount of P2O5 applied with LSM56.1 (Table 2) was usually more than double crop removal estimates (about 60 kg ha–1). Marginal increases were noted at LSM37.4 for both measures of available P (3 mg kg–1 for bicarb-P, 6 mg kg–1 for Bray-1 P). Therefore, at optimum LSM injection rates, small increases in soil test P could eventually be expected for LSM of composition similar to that used in our study. Initial soil test P values at the silt loam site (Table 3) approached levels above which a P index should be determined (30 mg kg–1; Hilborn and Stone, 2000) even though the field had no previous history of manure application. The dilemma of P buildup in soils receiving manure at N-based rates could be resolved by feeding low-phytate corn, which results in 42% less P in manure (Gollany et al., 2003). Soil P concentration did not differ between TD and INJ application methods, indicating that much of the apparent N loss with TD (as evidenced by lower grain yield) was likely due to volatilization rather than runoff.

Soil test K increased after 3 yr with LSM74.8 (100 to 130 mg kg–1), remained unchanged with LSM37.4, and was drawn down in control plots (105 to 82 mg kg–1). Therefore, at optimum LSM application rates, the amount of K applied (averaged 160 kg K2O ha–1, Table 2) likely balanced crop uptake, and at greater-than-optimum rates, K did not likely limit yields. Other soil properties such as micronutrient concentrations and SOM did not change with LSM application rate or method over time. A small reduction in pH (0.3 units) was noted at rates greater than LSM37.4 after 3 yr and may have been due to release of H ions during nitrification of manure NH4.


   SUMMARY AND CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
REFERENCES
 
Supplying corn nutrient requirements with sidedress-injected manure, applied using equipment that was commercially available by the end of the study period, produced above-average corn grain yields in both narrow- (51 cm) and wide- (75 cm) row corn. Field-scale (combine monitor) data served to validate and verify trends recorded at the smaller plot scale. The sidedress LSM injection rate required to optimize corn grain yield averaged 50 (plot-scale data) and 41 (field-scale data) m3 ha–1 over 4 yr. At near optimum application rate (LSM37.4), yield was 2 Mg ha–1 greater with INJ than TD (plot-scale data, more precise for quantification of treatment effects due to the absence of border effects). At greater application rates (LSM56.1), corn was less N-limited, and method of application had less effect on yield. Variable growth responses to TD and reduced yields relative to INJ were attributed mainly to variable N losses by NH3 volatilization with topdressing.

Three years of LSM injection at rates to optimize yield resulted in only marginal increases in soil test P, whereas soil P accumulated with above-optimal rates. Since greater TD application rates (LSM56.1) were required to produce yields equivalent to those with INJ (LSM37.4), topdressing with rates to optimize yield would lead to more rapid buildup of soil P than INJ. Soil test K did not change over time at optimal INJ rates, and so when supplying corn N requirements with sidedressed LSM, additional K would not likely be required unless initial soil levels were low or LSM composition differed from that used in this study. Carryover of N from previous LSM applications was minimal according to topsoil Ninorg measures presidedress, even where application rates exceeded crop demand. For example, only 6 kg NO3–N ha–1 accumulated before sidedress by 2002 following LSM93.5 in 2000 plus LSM74.8 in 2001. Some of the N applied in excess of uptake at above-optimal rates moved to subsoil layers, but much likely remained available for subsequent crops, as it was detected within 1 m of the soil surface.

Since yield response to sidedress LSM injection rate was a quadratic function for both plot- and field-scale data, and the PSNT was both well correlated with yield and correct in indicating whether additional N was needed most (88%) of the time, the PSNT can potentially be used to fine-tune sidedress LSM rates. We propose that in systems sidedressed with LSM, PSNT tables developed to generate Ninorg recommendations are also appropriate for predicting LSM application rates based on available N in manure. Further investigation to validate the practice is warranted.


   ACKNOWLEDGMENTS
 
Ontario Pork and AAFC's Matching Investment Initiative—financial support; K. Henning, A. More, and A. Dumayne—technical support; G. Milliken—statistical advice; Nuhn Industries, A & L Laboratories Canada, Green Lea Ag Centre, Van Raay Farms, T. Groenestege, Logan Tractor, Dekalb, and OMAF.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 SUMMARY AND CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Agron. J.Home page
M. H. Chantigny, D. A. Angers, G. Belanger, P. Rochette, N. Eriksen-Hamel, S. Bittman, K. Buckley, D. Masse, and M.-O. Gasser
Yield and Nutrient Export of Grain Corn Fertilized with Raw and Treated Liquid Swine Manure
Agron. J., August 11, 2008; 100(5): 1303 - 1309.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
B. R. Ball Coelho, R. C. Roy, E. Topp, and D. R. Lapen
Tile Water Quality following Liquid Swine Manure Application into Standing Corn
J. Environ. Qual., March 1, 2007; 36(2): 580 - 587.
[Abstract] [Full Text] [PDF]

Home page
 Soil Sci.Home page
B. R. Ball Coelho, R. C. Roy, and A. J. Bruin
Nitrogen Recovery and Partitioning with Different Rates and Methods of Sidedressed Manure
Soil Sci. Soc. Am. J., February 2, 2006; 70(2): 464 - 473.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ball Coelho, B. R.
Right arrow Articles by Bruin, A. J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Ball Coelho, B. R.
Right arrow Articles by Bruin, A. J.
Agricola
Right arrow Articles by Ball Coelho, B. R.
Right arrow Articles by Bruin, A. J.
Related Collections
Right arrow Soil Fertility and Productivity
Right arrow Animal Waste
Right arrow Best Management Practices
Right arrow Maize Management

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Crop Science Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome